back to article Aggressive HGST hurls flashy humdingers at online archiving

Holy Moly, HGST is getting ambitious. It's building an active archive platform product in competition with some of its OEMs and its aiming to rewrite server clustering with a flash fabric - oh and develop helium-filled disk drives - and shingled drives with its own slant - and thinking of Phase Change Memory chips with DIMM …

  1. Howverydare
    FAIL

    Pedant/Correction Alert

    Just a quick point:

    "NAND has its basic access latency - say 70ms, and PCIe adds 10ms to this whereas FlashDIMM adds perhaps 2ms - meaning no significant difference in overall data access latency; 80 ms vs 72 ms."

    If NAND plus PCIe has 80ms latency, how do Fusion IO and HGST's Virident acquisition deliver µs latencies on a PCIe card?

    1. Graham 24

      Re: Pedant/Correction Alert

      I suspect all the units are wrong - "DIMM's 2 ms" doesn't look right either. I'm fairly sure I can read more than 500 different locations from main RAM in 1 second!

      1. gazthejourno (Written by Reg staff)

        Re: Re: Pedant/Correction Alert

        Er, yes, about that. Brain/finger interface malfunction between ms and µs when abbrev'ing the units. Mea culpa.

    2. Anonymous Coward
      Anonymous Coward

      Re: Pedant/Correction Alert

      I can't even see how it makes sense if ms is being (ab)used as shorthand for microsecond.

  2. Alan Brown Silver badge

    even microseconds are too large.

    More like nano seconds.

    FWIW the access time of the best DRAM is still in the 60ns range - and hasn't really changed for 20 years. Other ram technologies have far lower latency (ZRAM is 2-4ns for example) but cost more or aren't yet commercially viable.(*)

    _worst_ case latency of PCIe 1.0 is 21ns, PCIe 2.0 is usually less than 10ns

    NAND read latency is variable, but even 70us would be inordinately long.

    (*) HP has been sampling memristors for a while. At 5-10ns latency I'd take 2Gb DIMMS _NOW_, given how much they'd speed up computational work needed here (the processors spend more time waiting for memory than they do crunching)

  3. Richard Boyce
    Thumb Down

    Shingles

    As a lay person, everything I hear about shingled drives gives me the willies. It may make sense in big data centres with lots of redundancy, but I will never buy such a drive. Until HAMR or better technology comes along, I'll make do with current PMR, even if that means buying extra drives.

    1. Tom Samplonius

      Re: Shingles

      "As a lay person, everything I hear about shingled drives gives me the willies... I will never buy such a drive."

      Don't worry. Such drives may never appear in the channel for you to buy. Millions of SMR drives have apparentlyl shipped, but apparently shipped directly to a cloud host (probably Amazon and Facebook). I can't even the find the model numbers for these drives.

      Given that the big 5 (Google, Facebook, Amazon, Apple and Microsoft) account for about half of all purchased computer hardware, more and more products will simply sold directly to the customer. It kind of sucks being a distie these days.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2019